A Deep Backward Scheme for Reflected BSDEs via Penalization : Approximation and Convergence Analysis

Abstract

We propose and analyze a deep backward numerical scheme for Markovian reflected backward stochastic differential equations with a generator independent of the control variable, based on the penalization method. The reflection constraint is replaced by penalized BSDEs, which are approximated backward in time by neural networks, with particular attention to the dependence of the errors on the penalization parameter λ. We derive quantitative error estimates for both the value process Y and the control process Z. A key ingredient is an L²-regularity estimate for the control component with explicit dependence on λ. Exploiting the monotonicity of the penalization term, we obtain stability estimates uniform in λ and prove uniform spatial Lipschitz bounds for both the penalized value functions and the discrete backward targets. Consequently, the GroupSort Lipschitz constraints and the associated neural-network approximation error can be controlled uniformly in λ. The global analysis separates penalization, time-discretization, forward, and neural-network approximation errors. Balancing the dominant terms yields a mean-square error of order O(|π|^1/3) for the value component, together with a corresponding estimate for a post-training reconstruction of Z. Numerical experiments with an explicit reflected BSDE and American option problems illustrate the scheme on one-dimensional and structured high-dimensional benchmarks.

Finite Element Approximation of a Hemivariational Inequality for Steady-State Heat Conduction: Double-Limit Convergence of Penalization and Discretization
Piotr Bartman-Szwarc, Anna Ochal, Domingo A. Tarzia
arXiv28 Sep 2026PreprintCC BY 4.0Full textAnalysis & PDEs Numerical analysis & scientific computingarXiv
Consistency and Convergence of the Backward-Euler Scheme for Stochastic Functional Differential Equations Driven by Fractional Brownian Motion
Alexander Abreu, Lisandro Fermin, Ernesto Mordecki +1
HAL29 Sep 2026PreprintCC BY-NC-ND 4.0Full textNumerical analysis & scientific computing ProbabilityHALarXiv
Backward and reflected backward stochastic differential equations under initial enlargement of Brownian filtrations
Youssef Ouknine, Badr Elmansouri
arXiv13 Oct 2026PreprintCC BY-SA 4.0Full textProbabilityarXiv
Conditional Reflected Backward Stochastic Differential Equations
Hanwu Li
arXiv21 Oct 2026PreprintarXiv licenceFull textProbabilityarXiv
Local W^2,1_p convergence of penalization for time-inconsistent stopping problems
Xiaodong Luo, Xiang Yu
arXiv4 Oct 2026PreprintarXiv licenceFull textOptimisation & controlarXiv

Cite

BibTeX RIS

BibTeX

@misc{essaky2026deep,
  title = {{A Deep Backward Scheme for Reflected BSDEs via Penalization : Approximation and Convergence Analysis}},
  author = {E. H. Essaky and M. A. Jalal and Y. Ouknine},
  howpublished = {arXiv},
  year = {2026},
  month = oct,
  doi = {10.48550/arxiv.2610.26420},
  eprint = {2610.26420},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2610.26420},
}

RIS

TY  - UNPB
TI  - A Deep Backward Scheme for Reflected BSDEs via Penalization : Approximation and Convergence Analysis
AU  - E. H. Essaky
AU  - M. A. Jalal
AU  - Y. Ouknine
PY  - 2026
DA  - 2026/10/19
PB  - arXiv
DO  - 10.48550/arxiv.2610.26420
UR  - https://arxiv.org/abs/2610.26420
AB  - We propose and analyze a deep backward numerical scheme for Markovian reflected backward stochastic differential equations with a generator independent of the control variable, based on the penalization method. The reflection constraint is replaced by penalized BSDEs, which are approximated backward in time by neural networks, with particular attention to the dependence of the errors on the penalization parameter λ. We derive quantitative error estimates for both the value process Y and the control process Z. A key ingredient is an L^2-regularity estimate for the control component with explicit dependence on λ. Exploiting the monotonicity of the penalization term, we obtain stability estimates uniform in λ and prove uniform spatial Lipschitz bounds for both the penalized value functions and the discrete backward targets. Consequently, the GroupSort Lipschitz constraints and the associated neural-network approximation error can be controlled uniformly in λ. The global analysis separates penalization, time-discretization, forward, and neural-network approximation errors. Balancing the dominant terms yields a mean-square error of order O(|π|^1/3) for the value component, together with a corresponding estimate for a post-training reconstruction of Z. Numerical experiments with an explicit reflected BSDE and American option problems illustrate the scheme on one-dimensional and structured high-dimensional benchmarks.
ER  -

Abstract

Related

Cite